13 instantiations of SentencePieceNormalizer
Microsoft.ML.Tokenizers (2)
Model\LlamaTokenizer.cs (1)
57SentencePieceNormalizer normalizer = new(
Model\SentencePieceTokenizer.cs (1)
76_normalizer = new SentencePieceNormalizer(modelProto.NormalizerSpec.RemoveExtraWhitespaces, AddDummyPrefix, EscapeWhiteSpaces, modelProto.TrainerSpec.TreatWhitespaceAsSuffix, specialTokens);
Microsoft.ML.Tokenizers.Tests (11)
LlamaTests.cs (11)
397SentencePieceNormalizer normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: false, addDummyPrefix: false, escapeWhiteSpaces: false, treatWhitespaceAsSuffix: false, specialTokens: null); 401normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: true, addDummyPrefix: false, escapeWhiteSpaces: false, treatWhitespaceAsSuffix: false, specialTokens: null); 405normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: true, addDummyPrefix: true, escapeWhiteSpaces: false, treatWhitespaceAsSuffix: false, specialTokens: null); 409normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: true, addDummyPrefix: true, escapeWhiteSpaces: true, treatWhitespaceAsSuffix: false, specialTokens: null); 413normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: false, addDummyPrefix: true, escapeWhiteSpaces: true, treatWhitespaceAsSuffix: false, specialTokens: null); 417normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: true, addDummyPrefix: true, escapeWhiteSpaces: true, treatWhitespaceAsSuffix: true, specialTokens: null); 421normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: true, addDummyPrefix: false, escapeWhiteSpaces: true, treatWhitespaceAsSuffix: true, specialTokens: null); 425normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: false, addDummyPrefix: true, escapeWhiteSpaces: true, treatWhitespaceAsSuffix: true, specialTokens: null); 429normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: false, addDummyPrefix: true, escapeWhiteSpaces: false, treatWhitespaceAsSuffix: true, specialTokens: null); 433normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: false, addDummyPrefix: true, escapeWhiteSpaces: true, treatWhitespaceAsSuffix: false, specialTokens: (_llamaPhi3Tokenizer as LlamaTokenizer)!.SpecialTokens); 441normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: false, addDummyPrefix: true, escapeWhiteSpaces: true, treatWhitespaceAsSuffix: true, specialTokens: (_llamaPhi3Tokenizer as LlamaTokenizer)!.SpecialTokens);
7 references to SentencePieceNormalizer
Microsoft.ML.Tokenizers (5)
Model\LlamaTokenizer.cs (1)
57SentencePieceNormalizer normalizer = new(
Model\SentencePieceTokenizer.cs (4)
1558char prefixSuffixChar = EscapeWhiteSpaces ? SentencePieceNormalizer.DummyPrefix : ' '; 1658Debug.Assert(sb[suffixIndex] == SentencePieceNormalizer.DummyPrefix); 1674return EscapeWhiteSpaces ? sb.ToString(SentencePieceNormalizer.DummyPrefix, ' ') : sb.ToString(); 1807char prefixSuffixChar = EscapeWhiteSpaces ? SentencePieceNormalizer.DummyPrefix : ' ';
Microsoft.ML.Tokenizers.Tests (2)
LlamaTests.cs (2)
75propertyInfo = typeof(SentencePieceNormalizer).GetProperty("TreatWhitespaceAsSuffix", BindingFlags.Instance | BindingFlags.NonPublic | BindingFlags.Public); 397SentencePieceNormalizer normalizer = new SentencePieceNormalizer(removeExtraWhiteSpaces: false, addDummyPrefix: false, escapeWhiteSpaces: false, treatWhitespaceAsSuffix: false, specialTokens: null);